Evaluation of the spatial variability of soil water content at the spatial resolution of SMAP data products : case studies in Italy and Morocco

Frequent and contiguous observations of soil water content such as the ones to be provided by SMAP arepotentially useful to improve distributed models of soil water balance. This requires matching of observations andmodel estimates provided both sample spatial patterns consistently. The spatial resolution of SMAP soil watercontent data products ranges from 3 km X 3 km to 40 km X 40 km. Even the highest spatial resolution may not besufficient to capture the spatial variability due to terrain, soil properties and precipitation.We have evaluated the SMAP spatial resolution against spatial variability of soil water content in two Mediterraneanlandscapes: a hilly area dominated by vineyards and olive orchards in Central Italy and a large irrigationschemes (Doukkala) in Morocco.The "Valle Telesina" is a 20,000 ha complex landscape located in South Italy in the Campania region, which has acomplex geology and geomorphology and it is characterised by an E-W elongated graben where the Calore riverflows. The main crops are grapevine (6,448 ha) and olive (3,390 ha). Soil information was mainly derived from anexisting soil map at 1:50 000 scale (Terribile et al., 1996). The area includes 47 SMUs (Soil Mapping Units) andabout 60 soil typological units (STUs). (Bonfante et al., 2011).In Doukkala, the soil water retention and unsaturated capillary conductivity were estimated from grain sizedistribution of a number of samples (22 pilot points, each one sampled in 3 horizons of 20cm), and combined witha soil map. The land use classification was carried out using a NDVI time series at high spatial resolution (LandsatTM and SPOT HRV).We have calculated soil water content for each soil unit in each area in response to several climate cases generatingdaily maps of soil water content at different depths.To reproduce spatial sampling by SMAP we have filtered these spatial patterns by calculating box averages withgrid sizes of 1 km X 1 km and 5 km X 5 km. We have repeated this procedure for soil water content in the 0 to 5cm and 0 to 10 cm depths. For each case we have compared the variance of filtered soil water content with theexpected accuracy of SMAP soil water content.The two areas are very different as regards morphology and soil formation. The Valle Telesina is characterizedby a very significant variability of soil hydrological properties leading to complex patterns in soil water content.Contrariwise, the soil properties estimated for all soil mapping units in the Dhoukkala collapse into just two pairsof water retention and hydraulic conductivity characteristics, leading to smoother patterns of soil water content.